Dear PyMVPA community,

I would like to get your feedback for an analysis I'm trying, I think I'm
in the right track but I don't want to miss something.

I have a block design in which I presented images belonging to 4 categories
and I took 6 acquisitions. The thing is that I scanned dogs and humans.

I want to test whether both species represent the categories in a similar
way. I took the coordinates of a voxel of interest in the human brain, took
the voxels withing a sphere around it and calculated the euclidean distance
between blocks. Then I took this "goal distance vector" and calculated the
spearman correlation of similar vectors but obtained in a sphere around
each voxel a dog's brain. I repeated this procedure for each human and each
dog. Thus getting correlation maps for each dog x human. Then I threshold
the maps and binarized the results. I finally added the maps and creating a
group map which tells me how many correlations were above chance on each

To test wheter it is significant or not, I took random coordinates in the
human brain and repeated the process so I can get a distribution of how
many false positive correlations I can get in a map, and then I plan to use
this distribution to have a statistical threshold.

Does it make sense to you?

Sorry for the long explanation :s


Pkg-ExpPsy-PyMVPA mailing list

Reply via email to